Abstract: Poor convergence to concave shapes is a main limitation of snakes as a standard segmentation and shape modelling technique. The gradient of the external energy of the snake represents a force that pushes the snake into concave regions, as its internal energy increases when new inflexion points are created. In spite of the improvement of the external energy by the gradient vector flow technique, highly non convex shapes can not be obtained, yet. In the present paper, we develop a new external energy based on the geometry of the curve to be modelled. By tracking back the deformation of a curve that evolves by minimum curvature flow, we construct a distance map that encapsulates the natural way of adapting to non convex shapes. The gradient of this map, which we call curvature vector flow (CVF), is capable of attracting a snake towards any contour, whatever its geometry. Our experiments show that, any initial snake condition converges to the curve to be modelled in optimal time.

Abstract: In this paper we propose a new topology based metric for 2D vector fields. This metric is based on the concept of feature flow fields. We show that it incorporates both the characteristics and the local distribution of the critical points while keeping the computing time reasonably small even for topologically complex vector fields. Finally, we apply the metric to track the topological behavior in a timedependent vector field, and to evaluate a smoothing procedure on a noisy steady vector field.

Abstract: The paper presents an approach for making real-time three-dimensional vector flow imaging. Synthetic aperture data acquisition is used, and the data is beamformed along the flow direction to yield signals usable for flow estimation. The signals are cross-related to determine the shift in position and thereby velocity. The data can be beamformed after reception in any direction and any vectorial velocity can be found. More than 60 independent velocity volumes can be made per second with this approach. A 3 MHz 2D matrix transducer consisting of 64 /spl times/ 64 elements with /spl lambda//2 pitch are used. The emissions are done using 16 /spl times/ 16 = 256 elements at a time and the received signals from the same elements are sampled. Access to the individual elements is done through 16-to-1 multiplexing, so that only a 256 channels transmitting and receiving system are needed. The method has been investigated using Field II. Parabolic flow in a 10 mm radius vessel inclined at 60 degrees to the acoustical axis of the transducer was simulated. The mean standard deviation of he estimates was 0.0098 m/s over the whole vessel cross-section, which is 3.3% relative to the peak velocity. The bias was 0.023 m/s (7.5%). False peaks were found mainly at the edges of the vessel due to the echo-cancelling, and the probability of false detection was 2.2%.

Abstract: A nonlinear surface registration algorithm of thoracic/abdominal structures segmented from CT and PET volumes is presented. The aim of this work is to develop a method that can provide an initial estimate of the elastic deformation between the images, so that MI-based techniques can be successfully applied. To perform the matching, a B-spline Free Form Deformation (FFD) model has been chosen. Hierarchical structure segmentation and rigid registration are applied to initialize the nonlinear surface registration phase. Two different approaches to optimize the warp are tested: an iterative gradient descent technique based on local gradient estimations over the grid of control points; and an original optimization based on Gradient Vector Flow (GVF) computed on the CT image. Finally, we evaluate our results, using an Iterative Closest Point (ICP) rigid registration algorithm as a reference to compare both approaches.

Abstract: In this paper we propose a new topology based metric for 2D vector fields. This metric is based on the concept of feature flow fields. We show that it incorporates both the characteristics and the local distribution of the critical points while keeping the computing time reasonably small even for topologically complex vector fields. Finally, we apply the metric to track the topological behavior in a timedependent vector field, and to evaluate a smoothing procedure on a noisy steady vector field.

Abstract: A solution to the problem of model-based image segmentation by integrating prior color and texture information into a parametric active contour model is proposed in this paper. The vector flow field technique is adopted as the computational scheme because of its large effective capture range. Color and texture priors are modeled in a uniform way to generate a potential field and its associated vector flow field for the repetitive application of the gradient vector field (GVF). The initialization of the snake is decided by studying the shape properties of the potential field. The shape information can be utilized to avoid unnecessary initializations caused by disturbing background similar color and texture features to the model object. The potential field is also valuable for determining the weights for integrating the color and texture based vector flow field and the edge based vector flow field. With the proper integration weights, the snake evolved on the combined vector flow field can bypass strong edges in the background and stop correctly on the object's boundary. Experimental results are shown on real scenes with complex backgrounds.

Abstract: A novel method is introduced to force a geometric-based snake be more tolerant towards weak edges and noise in images. The method integrates gradient flow forces with region constraints obtained from diffused region segmentation forces. The diffusion is obtained from the region map vector flow field. This extra region force gives the snake a global view of the boundary information within the image. We present results on both graylevel and colour images.

Abstract: Using the active contours model (ACM) to estimate the cardiac motion, the new concept of generalized fuzzy gradient vector flow (GFGVF) is presented in this paper. The GFGVF is refered as a component of external force and associated with optical flow field (OFF) to build a set of Snake equations. After the GFGVF and OFF are accurately calculated respectively, the initial outline can gradually approach to the region of interest (ROI) edges in the images under the constraint of Snake equations and track the ROI from frame to frame. Under some constrained conditions, the motion states of some feature points in the edge of ROI can be found by the Maximum a Posteriori Probability (MAP) during a period of cardiac motion. Then, the motion estimation is well optimized. Another, the coefficients in the set of ACM equations can be found using the prior information, which avoids giving them by experience and improves the capability of edge tracing of ROI. By a period of motion tracking for CT and MR cardiac image sequences, the experiments show that the method can robustly simulate the motion of the cardiac left ventricle (LV) and left atria (LA) , moreover, the simulation result of them using GFGVF is obviously better than the one using GVF.

Abstract: Face detection is an important research area having wide applications in man-machine interface, visual surveillance and face recognition. We have proposed an algorithm for face detection using Gradient Vector Flow in gray level images which overcomes the problem for localization and initialization. The algorithm has been tested on various face databases and the result shows an accuracy of 97% with invariance to pose and orientation.

Abstract: Recording rolling leukocyte velocities from intravital microscopic video imagery is important to inflammation research and drug validation. Since manual tracking is excessively time consuming, an automated method is desired. This paper illustrates an active contour based automated tracking method, where we propose a novel external force to guide the active contour by taking the flow direction into account. The construction of the proposed force field, referred to as motion gradient vector flow (MGVF), is accomplished by minimizing an energy functional involving the motion direction, and the image gradient magnitude. The tracking experiments demonstrate that MGVF can be used to track both slow and fast leukocytes, whereas the basic gradient vector flow (GVF) is suitable for tracking slow leukocytes.

Abstract: Several aspects in the numerical analysis system of electroencephalography (EEG) are discussed in this paper. Based on improved vector flow snakes and dynamic template method, the boundaries of different region in MRI image are extracted. Using the boundary element method, forward computation of EEG is investigated and acceptable results are obtained. The paper also proposes a method based on LORETA-FOCUSS to estimate the current density inside the brain with high spatial resolution. From simulated experiments, it is turned out that the proposed techniques yield satisfactory results.

Abstract: The standardgeometric orgeodesic active contourisa powerful segmentation method, yet it is susceptible to weak edges and image noise. We propose a new region-aided, geometric, colour active contour that integrates gradient flow forces with region constraints. These constraints are composed of image region vector flow forces obtained through the diffusion of the region segmentation map. The extra region force gives the snake a global view of the boundary information within the image which, along with the local gradient flow, helps detect fuzzy boundaries and overcome noisy regions. The partial differential equation (PDE) resulting from this integration of image gradient flow and diffused region flow is implemented using the level set approach.

Abstract: Active contours, or snakes, are used extensively in digital image analysis and computer vision applications, particularly to locate object boundaries, which is very useful in medical image processing, such as MRI and CT images. However, due to speckle noises, weak edges and tissue related textures in ultrasound images, the conventional active contour models usually can not obtain satisfying contour tracking results. Based on gradient vector flow(GVF) active contour model, this paper introduced a conception of band limitation and applied this active contour model into ultrasound image segmentation. The experimental results show that this algorithm limits the influence of false edge and noise disturbance and obtains desired segmentation results for ultrasound and ultrasound serial images.

Abstract: In this paper, an active contour model based on the gradient vector flow of the gray-level and the motion similarity measure (MGVF-Snake) is introduced after analyzing the performance of the traditional Snake and Snake based on gradient vector flow (GVF-Snake), the algorithm is proposed to extract and track the Video Object (VO) automatically. The MGVF-Snake overcomes the shortcoming of the GVF-Snake that could not fine the VO contour precisely in the complex background. In allusion to the problem that the traditional GVF-Snake easily makes mistake when tracing the VO moving rapidly, the scheme takes advantage of the redundancy and makes the tracking more accurately and rapidly by adjusting the previous VO contour with the motion vector as the current initial contour. The algorithm is validated with the video sequences, the results of the experimentation show that it not only extract VO contour automatically, but also track accurately.

Abstract: We present the results of two studies investigating the optimal aperture configuration for maximized lateral blood flow velocity estimation using Heterodyned Spatial Quadrature. Our objective was to determine the maximum velocities that can be estimated at Doppler angles of 90 degrees and 60 degrees with a bias of less than 5% for both uniform scatterer motion in a tissue-mimicking phantom and blood-mimicking fluid circulated through a wall-less vessel flow phantom. Constant flow rates ranging from 3.0 to 18.0 ml/sec were applied in the flow phantom, producing expected peak velocities of 15.0 to 89.8 cm/sec under laminar flow conditions. Velocity estimates were obtained at each flow rate using 256 trials, with each trial consisting of an ensemble of 32 vectors. For an f/1 receive geometry with bi-lobed Hamming apodization, all peak flow velocities tested were estimated to within 5% of their expected values for both 90 degree and 60 degree Doppler angles. An f/2 receive geometry featuring bi-lobed Blackman apodization generally provided accurate lateral velocity estimates up to 71.9 cm/sec for a Doppler angle of 90 degrees, and accurate lateral component estimates up to 50.1 cm/sec for a 60 degree Doppler angle. The implications of these findings will be discussed.